2026-03-23 東北大学
東北大学の研究グループは、AR/MR機器向けに小型バーチャルキーボードでも快適に文字入力できる新手法「FanType」を開発した。利用者の入力意図をアルゴリズムで推定し、候補文字を扇形に提示することで、親指の小さな動きだけで高精度な入力を可能にした。従来の空中入力の課題であった誤入力や操作負担を軽減し、混雑した環境でも省スペースで効率的な操作を実現。AR/MR機器の実用化を支える入力インタフェース技術として期待される。
図1. 混雑した公共空間を想定した環境でAR機器を装着し、限られたスペースの中で親指操作による文字入力を行う様子。
＜関連情報＞
- https://www.tohoku.ac.jp/japanese/2026/03/press20260323-03-armr.html
- https://www.tohoku.ac.jp/japanese/newimg/pressimg/tohokuuniv-press20260323_03web_ARMR.pdf
- https://ieeexplore.ieee.org/document/11474501
FanType：小型XRキーボードでのテキスト入力向けに、意図を推論する扇形の親指インターフェース FanType: Intention-Inferring Fan-shaped Thumb Interface for Text Entry on Small XR Keyboards
Guanghan Zhao, Louis Teys, Gyeonghwan Yang, Shengdong Zhao, and Yoshifumi Kitamura
IEEE Xplore
口頭発表: 2026 IEEE Conference Virtual Reality and 3D User Interfaces (VR), Daegu, Korea, 2026年3月23日
IEEE Transactions on Visualization and Computer Graphics Date of Publication: 03 April 2026
DOI:https://doi.org/10.1109/TVCG.2026.3680595
Abstract
In this paper, we present FanType, a text entry technique that enables efficient typing on small virtual keyboards in extended reality (XR), closely resembling thumb-based typing on smartphones and tablets. While virtual environments theoretically provide unlimited space, practical XR usage often occurs in mobile and spatially constrained contexts, such as when using augmented reality (AR) headsets while walking or in confined environments, where large virtual keyboards are impractical, as they occupy substantial visual space and interfere with other virtual or real content. To address this, FanType integrates a fan-shaped interface that groups keys within the thumb’s reach and leverages a lift-up gesture for disambiguation, thereby reducing hand movement while preserving natural typing speed. Additionally, an intention-inferring mechanism involving tailored algorithms is implemented to counterbalance touch bias and further ensure input accuracy. A user study with 22 participants evaluated the method on tablet- and smartphone-size portrait keyboards, comparing it with a baseline (touch-only) and a state-of-the-art mid-air technique. Results provide insights into the performance of such a design and highlight the challenges of designing typing methods for small virtual keyboards
